Explicit Subcodes of Reed-Solomon Codes that Efficiently Achieve List Decoding Capacity

Amit Berman,Yaron Shany,Itzhak Tamo

2024-01-27

Abstract:In this paper, we introduce a novel explicit family of subcodes of Reed-Solomon (RS) codes that efficiently achieve list decoding capacity with a constant output list size. Our approach builds upon the idea of large linear subcodes of RS codes evaluated on a subfield, similar to the method employed by Guruswami and Xing (STOC 2013). However, our approach diverges by leveraging the idea of {\it permuted product codes}, thereby simplifying the construction by avoiding the need of {\it subspace designs}.

Information Theory

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to construct a new family of explicit sub - codes. These sub - codes are subsets of Reed - Solomon (RS) codes and can efficiently reach the list - decoding capacity while maintaining a constant - sized output list. Specifically, the paper proposes a method through permuted product codes to achieve this goal, which simplifies the previous construction methods and avoids the complex process of designing sub - spaces.

### Background and Problem Description of the Paper

In coding theory, error - correcting codes are used to reliably transmit data over noisy channels. A code \(C\subseteq\Sigma^{n}\) (where \(\Sigma\) is the alphabet and \(n\) is the length) is usually a set containing \(|\Sigma|^{k}\) codewords, where \(k < n\). We call the rate of this code \(R:=\frac{k}{n}\). A fundamental goal in coding theory is to maximize the minimum Hamming distance \(d\) between codewords given the rate \(R\).

For a code with a minimum distance \(d\) and a normalized distance \(\delta:=\frac{d}{n}\), if the received codeword has errors in less than \(\frac{\delta}{2}\) positions, the transmitted codeword can be uniquely determined. However, when the error proportion exceeds \(\frac{\delta}{2}\), the transmitted codeword usually cannot be uniquely determined. Since the Singleton bound shows that \(\delta\leq1 - R\), the maximum possible unique decoding radius for a code with rate \(R\) is \(\frac{1 - R}{2}\), which can be achieved by Reed - Solomon (RS) codes.

To go beyond the unique decoding radius, the decoder must be allowed to output a list of potentially transmitted codewords. A code \(C\) is said to be \((\rho, L)\)-list - decoded if for any received word \(w\), there are at most \(L\) codewords \(c\in C\) such that \(c\) differs from \(w\) in no more than a proportion \(\rho\) of positions.

### Main Contributions of the Paper

1. **Constructing New Explicit Sub - Codes**: The paper proposes a new family of explicit sub - codes. These sub - codes are subsets of RS codes and can efficiently reach the list - decoding capacity. Specifically, these sub - codes are constructed by the method of permuted product codes, avoiding the complex process of designing sub - spaces.

2. **Simplifying the Construction Method**: The method in the paper is based on known tensor product codes and constructs codewords by specific cyclic shifts. This construction method is simple and easy to implement.

3. **Using Two Affine Transformations**: The paper also proposes a method of defining codes using two affine transformations with relatively prime orders, which enables the code length to reach the size of the finite field without requiring the finite field to be prime.

### Technical Details

- **Permuted Product Codes**: By constructing the tensor product of two RS codes and performing specific cyclic shifts on the rows, we obtain permuted product codes where each codeword column is regarded as a single coordinate.
- **Affine Transformations**: Use two affine transformations with relatively prime orders to generate codewords, which extends the previous method of using a single affine transformation.
- **List - Decoding Algorithm**: The paper proposes a linear - algebraic list - decoding algorithm that can efficiently decode near capacity.

### Summary

The main contribution of the paper is to provide a new, simple construction method that enables sub - codes of RS codes to efficiently reach the list - decoding capacity while maintaining a constant - sized output list. This method not only simplifies the construction process but also extends previous work, allowing the code length to reach the size of the finite field.